The spectral calibration of the PSPC unfortunately remains incomplete. MPE and the USRSDC have been working in collaboration to understand residual systematic effects in spectral fitting. In the course of our investigations we have found a number of unexpected complications which have made the task considerably more difficult.
We have determined that the effective gain across the PSPC is not uniform, and has a time dependency (see the document cal_ros_95_003 in the legacy anonymous ftp directory /caldb/docs/rosat). Using the data from the on-board Al K-alpha calibration source collected during regularly scheduled calibration intervals, the mean channel of the monochromatic distribution has been determined to vary by roughly +/-3% over the field. Of special note is a gain depression in the center of the field at the location where most sources are targeted. This gain depression can be as much as 4.5%, and preliminary results indicate that this effect has, in general, gotten worse with time, but not in a linear or monotonic manner. Since SASS corrects for spatial gain variations using preflight calibration data and for temporal gain variations using a global measure of the gain (i.e., the average gain across the detector), it cannot correct for these specific discrepancies.
A nonlinearity in the pulse height to energy scale, which is apparently a function of absolute gain of the detector, was first identified in astrophysical spectra and then verified by MPE using post-launch calibration data collected at Neureid. The effect is greatest in the 0.5-1.0 keV energy range where the output pulse-height channel for a given energy X-ray is lower than expected. (See document pspc_gain_memo in the legacy anonymous ftp directory /rosat/doc/mpe_memos.)
Two further uncertainties in the calibration are at the extremes of the response. At low energies there are discrepancies in the response between high- and low-gain data which are the greatest for particularly soft sources such as white dwarfs.
The residual systematics at low energies are particularly hard to address because of the coupling between effective areas (the transmissions of the PSPC windows were measured with a lowest energy of 183 eV) and the fundamental spectral response (again usefully measured at a minimum energy of only 183 eV). There is additional concern that the window has suffered some ablation with exposure to X-rays, and therefore may not have a constant thickness over the lifetime of the detector. The thickness varies somewhat spatially as well. The uncertainty at high energies is being addressed using Neureid ground calibration data.
Investigations are following several different lines. SASS software is being studied to eliminate an error in the event pulse-height correction algorithms which has been responsible for some of the residual spatial gain variations discussed in the document cal_ros_95_003 (specifically, greater gain variation in PI channels than in PHA channels). A correction algorithm which will operate on PI data is being developed to produce more uniformity across the detector and to correct for spectral nonlinearities. The major difficulty in refining the calibration is that there are no good astrophysical "calibration" sources. Systematic errors in the calibration are difficult to probe when the systematics of the calibration sources are model dependent.
It is important to note, however, that in general the calibration is relatively good. In most cases for specific sources, counting statistics will dominate many if not all of the residual systematic effects. Errors in the gain correction and gain linearity are both on the few percent level. The calibration errors at low energies are harder to gauge. Two rules of thumb for testing for systematic effects when considering the reality of any spectral result are to 1) fit the data with the two available matricies and 2) shift the gain scale by a few percent. If the observed effect is persistent through these tests, the likelihood that it is real is increased.
Select another edition of the ROSAT Newsletter
Please use the Feedback link if you have questions on ROSAT.